The present invention relates to a procedure as defined in the preamble of claim 1 and to an apparatus as defined in the preamble of claim 7 for the deceleration of an elevator.

According to various elevator regulations, an elevator must be able to stop at a landing with a certain accuracy. The re¬ quired tolerance is typically of the order of ±5 mm, which is easily attained by modern elevators. However, a greater stop¬ ping precision is aimed at, because the stopping accuracy is also regarded as a measure of quality of the elevator. Moreo¬ ver, the co-operation between certain parts of the elevator equipment, such as the car door and the landing door, is bet- ter in an elevator capable of accurate stopping.

The determination of elevator position is implemented using pulse tachometers mounted in conjunction with the machinery and giving pulse counts that are directly proportional to the revolutions performed by the machine. Another device used for the determination of elevator position is a tachometer which produces an analog voltage proportional to the elevator speed and whose output voltage is converted into a pulse train in which the pulse frequency is proportional to the speed and the pulse count to the distance covered by the elevator. How¬ ever, in both tachometer types, the distance calculated from the pulse count is not quite accurate because the elevator is driven by means of the friction between the elevator ropes and the traction sheave. The distance calculated from the ta- chometer pulses contains a small error, because there occurs a slight movement of the elevator ropes relative to the trac¬ tion sheave. Although the error in the calculated distance is not large, usually only a few millimeters, an objective in modern elevator technology is to eliminate even this small error.

Various solutions have been proposed to solve this problem, e.g. by updating the pulse counts representing elevator posi¬ tion at each floor, as is done in specification US 4,493,399. In some elevators two tachometers, an analog tachometer and a pulse tachometer, are used, together or separately. Another solution used to indicate elevator position is to provide the shaft or car with code reading devices producing accurate po¬ sition data.

The behavior of an elevator is also controlled by factors re¬ lating to passenger comfort, such as e.g. acceleration, de¬ celeration and changes in them, which, though in fact irrele¬ vant to the problem of determining elevator position, impose certain edge conditions regarding elevator control .

The object of the present invention is to integrate the ac¬ celeration and deceleration of an elevator and their changes as well as the calculation of elevator position with the ele¬ vator control so as to achieve a good stopping accuracy and a desired level of travelling comfort when the elevator is be¬ ing stopped at a floor.

To achieve the objects mentioned above, the procedure of the invention is characterized by what is presented in the char- acterization part of claim 1. The apparatus of the invention is characterized by what is said in the characterization part of claim 7. Other embodiments of the invention are character¬ ized by the features presented in the other claims.

When the procedure of the invention is applied, the elevator will have maximal performance characteristics, such as a high stopping accuracy and a comfortable travelling behavior within the framework of given performance parameters, such as acceleration, deceleration and the change in acceleration and deceleration (jerk) . The procedure of the invention obviates the need to carry out adjustments of deceleration elements

during installation.

According to the solution presented, the required decelera¬ tion is determined continuously on the basis of the remaining distance and the elevator is accordingly brought smoothly to the landing. The deceleration is changed continuously towards a point at which, using a calculated jerk, the speed, decel¬ eration and remaining distance become zero.

In the following, the invention is described by the aid of an embodiment by referring to the drawings, in which

- Fig. 1 presents an elevator environment according to the invention, - Fig. 2 represents correct operation of an elevator when reaching a target floor,

- Fig. 3 represents a case of premature stopping,

- Fig. 4 represents a case of belated stopping,

- Fig. 5 represents correction of premature stopping, - Fig. 6 illustrates the interconnections between decelera¬ tion, velocity and position in the solution of the inven¬ tion,

- Fig. 7 presents a block diagram of the deceleration phase of an elevator, - Fig. 8 represents the process of defining a reference value during the deceleration phase, and

- Fig. 9 represents the process of defining the change of deceleration during the final round-off.

The elevator car 2 (Fig. 1) is suspended on a hoisting rope 4 which is passed around the traction sheave 6, with a counter¬ weight 8 attached to the other end of the rope. To move the elevator, the traction sheave 6 is rotated by means of an elevator motor 10 coupled to its shaft and controlled by a control gear 12. The control gear 12 comprises a frequency converter which, in accordance with control signals obtained

from a control unit 14, converts the electricity supplied from a network 16 into the voltage and frequency required for the elevator drive. The control unit 14 sends the control pulses to the solid state switches of the frequency con- verter. The control unit 14 receives a frequency and ampli¬ tude reference via conductor 22 from the regulating and cal¬ culating unit 24 of the elevator or, more specifically, from a controller 26. To generate speed feedback, a tacho- generator 18 is connected to the traction sheave shaft either directly or via a belt to produce a tacho-voltage propor¬ tional to the speed of rotation.

The tacho-voltage proportional to the speed of the elevator motor is passed to an analog/digital converter, which gives the motor speed as a digital quantity consistent with the SI system, which is fed into the regulating and calculating unit 24 of the elevator. Stored in this unit 24 are nominal val¬ ues, selected for the elevator drive, for the jerks 21, ac¬ celeration 23, drive speed 25 during the constant-velocity stage and other parameters 27, such as coefficients determin¬ ing the margin by which the acceleration or jerk may be higher or lower than its nominal value. From a flag 34 mounted in the elevator shaft, the system obtains data indi¬ cating the elevator position in the vicinity of a landing, and this data is taken via conductor 36 to the regulating and calculating unit 24. In a manner to be described later on, a speed reference unit 29 calculates from the above-mentioned quantities a speed reference for the elevator at different phases of the movement of the elevator car so that, after leaving a landing, the elevator car is optimally accelerated to the highest possible drive speed and especially stopped smoothly exactly at the target floor. The distance form the floor as required for the calculation is defined as a time integral of the speed signal. The speed reference obtained from unit 29 together with the speed signal is fed into a discriminating element 35 and the output 37 of the discrimi-

nating element is fed into the controller 26, known itself, which contains a PI controller and produces the frequency and amplitude reference for the control unit 14. In a preferred embodiment of the invention, the control is implemented as a software based solution, but the invention can also be imple¬ mented using components performing the corresponding func¬ tions .

At point 48, when the elevator car reaches the deceleration point of the target floor, reduction of the speed reference is started, first at the jerk3 stage with a changing decel¬ eration using a nominal jerk up to point 50, then with con¬ stant deceleration to point 52 and finally with a changing deceleration during the final round-off to point 40. If de- celeration is started from the nominal speed using nominal deceleration and a nominal jerk, the deceleration point must be exactly right to enable the elevator to stop exactly at the floor level of the target floor. In this case the drive speed curve corresponds to the drive speed curve for accel- eration described above. Fig. 2 represents a case like this. In the situation represented by Fig. 3, the deceleration point 48' has been calculated as being located at a longer distance from the floor level than it actually is. With nomi¬ nal jerks and nominal deceleration, the elevator stops before the floor level at point 40' while the speed is changed as indicated by the broken line 54. Correspondingly, in the case illustrated by Fig. 4, the deceleration point has been calcu¬ lated as being located at point 48'' and consequently the elevator speed is decelerated as indicated by curve 56 and the elevator stops at point 40'' .

If the driving distance is so short that the nominal speed cannot be reached, then a transition is made from the con¬ stant acceleration phase in Fig. 2, 3 and 4 via a change of acceleration directly to the constant deceleration phase. The durations of the constant acceleration and deceleration

phases and, correspondingly, the maximum drive speed change in accordance with the driving distance. This has no effect on the deceleration procedure, which will be described later on, but the system functions in the same way even in this situation after the onset of constant deceleration.

Fig. 5 shows the deceleration phase of the situation repre¬ sented by Fig. 3 in a magnified view in order that the con¬ trol procedure of the invention can be described more explic- itly. The deceleration as provided by the invention as well as the speed reference and the final round-off or rate of change of deceleration before stopping are determined in the manner illustrated by the block diagrams in Fig. 7, 8 and 9. The calculation procedure is performed by the speed reference calculating unit and the speed reference obtained as a result is fed into the control unit 14. The elevator now decelerates at an optimal rate and so that, at the instant of stopping, the elevator is at the level of the target floor and its speed and deceleration are zero. Thus, the elevator reaches the target floor as quickly as possible from the deceleration point to the floor level and the deceleration occurs smoothly without any abrupt changes in speed or deceleration.

At the start of the deceleration phase, the speed reference is altered by the amount of the nominal jerk, and the decel¬ eration and speed are calculated according to the following equations

a _de = J- t _r

^β ώ = V ref

2 , where

- t _r is the rounding time of the speed curve starting from the deceleration point with differential steps dt starting from the value dt,

- a _de is the deceleration reference, which is changed by the amount of the nominal jerk,

- J is the nominal jerk, which has been selected as a de¬ fault value for acceleration changes at start and at the end of constant acceleration, jerkl, jerk2 and jerk3,

- a _dl is a deceleration value as calculated from the remain- ing distance to the floor level,

- d is the distance to the floor level of the target floor,

- d _x is the travel distance required for the final round¬ off, i.e. the additional distance to be traveled because of the final round-off in addition to the distance that would be traveled if the elevator were decelerated with constant deceleration to the target floor. d _x is calcu¬ lated using a pre-selected jerk value (=nominal jerk) .

The deceleration quantities a _de and a _dl are calculated and their values are compared with each other. The transition to constant deceleration is subject to the following require¬ ment: a _de ≥ a _dl .

If this condition for a transition to constant deceleration is not fulfilled, a new speed reference for the changing de¬ celeration phase will be calculated at the next instant fol¬ lowing the previous calculation after the lapse of the dif¬ ferential step dt.

During the constant deceleration phase, the speed reference is reduced in accordance with the block diagram in Fig. 7. According to the invention, during the constant deceleration phase the system is trying to find a point where the final deceleration can be started with the allowed jerk, i.e. where the transition to the final round-off on the speed reference curve is to occur. When this point (corresponding to point 52

in Fig. 2 - 5) is found, the deceleration is changed from then on by a constant jerk and the acceleration and speed references are changed accordingly, with the result that the acceleration, speed and distance from the target floor reach zero value at the same instant. Fig. 6 shows how the speed reference v _re£ , the distance d and the deceleration reference a _di , calculated using the distance and the nominal jerk, and correspondingly a _de , change as functions of time. In block 60, a proposed future value of the speed reference is calculated by reducing the value of the speed reference by the amount of a _de *dt. Based on the remaining distance, a new a _d i value (block 62) is calculated according to a formula to be pre¬ sented later on in connection with Fig. 8. If the difference between the deceleration reference a _de and the deceleration a _di calculated on the basis of the distance exceeds the al¬ lowed deceleration deviation Δa=J*dt, the deceleration a _de will be corrected by Δa (blocks 64, 65) . Correspondingly, the deceleration is corrected by Δa if the above-mentioned dif¬ ference is smaller than -Δa (blocks 64 and 66) or, if the difference is smaller, the current deceleration a _de is main¬ tained. In this way, the speed reference is made to follow the deceleration, which has been calculated on the basis of the remaining distance to the floor level, or if the devia¬ tion exceeds Δa, the deceleration reference can be made to approach the deceleration calculated on the basis of the dis¬ tance in steps of Δa, so the change will take place without any large jerks. Fig. 6 shows the change in a _di and a _de at the beginning of deceleration towards their point of coinci¬ dence at instant ti, which is when the constant deceleration phase begins. For example, when position correction (vane edge, flag) occurs during deceleration, the sudden change in the position data changes the deceleration reference, by means of which it is possible to produce a smooth round-off in the speed curve. The deceleration reference a _de is now changed in steps towards the deceleration reference a _di cal-

culated on the basis of distance until they are equal. The changes in the distance, deceleration and speed reference can be observed at point t ₂ in Fig. 6, at which a stepwise dis¬ tance correction is made. The deceleration a _di calculated on the basis of the distance changes in a stepwise manner (broken line) , while the deceleration reference or the decel¬ eration a _de (solid line) corresponding to the speed reference changes more slowly. In the curve of the speed reference v _ref , the change is visible as an almost imperceptible change in the slope. In block 68, based on the new deceleration refer¬ ence, a new speed reference v _ref is calculated, whereupon the value of the change J4 of deceleration for the final round¬ off is calculated (block 70) , which is presented in greater detail in Fig. 9. If the condition for starting the final round-off exists (block 72) , the final round-off phase will be activated. If not, action will be restarted from block 60 and a new speed reference will be calculated.

The procedure depicted in Fig. 8 is used to determine the speed reference during deceleration. In selection block 80 a check is made to see if the elevator is close to the floor level and if the flag has been detected. If there is no flag data and the distance calculation indicates that the elevator is at a distance below 150 mm from the floor (block 82), then an estimate d _err of position or distance error is generated, to be used in the deceleration value a _di (block 88) calcu¬ lated on the basis of distance. The position error d _err is in¬ creased by the step v _rβf *dt (block 84) and this correction is made on each calculation cycle when the position counter in- dicates that the flag should have been reached but the flag has not been detected. In this way, the position data is cor¬ rected in advance towards the probable absolute position. Us¬ ing the speed reference and the deceleration reference, a proposed new speed reference v=v _ref -a _de *dt (block 86) is cal- culated. Based on an ascertained or corrected estimate, the deceleration is calculated, using the distance to the target

floor, as a _dl =v ² / (2* (d+d _err -d _x ) ) , where d _x is the distance re¬ quired for the final round-off when the nominal jerk value is used (block 88) . The maximum allowed deceleration value a^ _x , for which a suitable value is kι*nominal deceleration (for instance, kι=1.3), is calculated (block 90), whereupon in block 92 a check is performed to see if the deceleration value a _dl calculated on the basis of distance exceeds the maximum deceleration value, to which the deceleration is lim¬ ited (block 94) if the maximum deceleration is exceeded. If the difference a _dlf£ (block 96) between the a _dl based on dis¬ tance and the deceleration reference a _de is larger than the reference value (=J*dt, where J is the default jerk value) and the deceleration reference is below the maximum, then the deceleration reference will be increased by the value J*dt (blocks 98 and 100) . If the condition applied in block 98 is not valid, then a check is made (block 104) to see if the de¬ celeration reference is above the minimum allowed decelera¬ tion reference a _mιn =k ₂ *nominal acceleration (preferably k ₂ = 0.7) (block 102) and if the difference a _dl-- between the a _dl calculated on the basis of distance and the deceleration ref¬ erence a _de is less than the reference value (=-J*dt) , and in a positive case the deceleration reference a _de is reduced by the amount of J*dt. Using deceleration references corrected in blocks 100 or 106 or, if no changes are allowed, an un- changed deceleration reference, a new speed reference value is calculated (block 108) . Finally the speed reference is checked to ensure that it is not below zero (blocks 110 and 112) and a jerk value J4 for the final round¬ off is calculated (block 114) . If the jerk has a non-zero value, the final round-off will be started using the calcu¬ lated jerk value, producing a speed curve with a final round¬ off determined by the selected jerk. If the jerk is zero, the procedure will continue with a repeated speed reference cal¬ culation. For the calculation of the jerk J4 for the final round-off in the manner provided by the invention, there are two edge con-

ditions, one for a case where the elevator is going to stop at a level past the floor and the other for a case where the elevator is stopping at a level before the floor. In addi¬ tion, there are conditions for calculating the jerk in a nor- mal case. If the initial data have not been defined (block 120) , then a minimum deceleration a _mn , a speed limit v _slm and a distance limit d _slιm (124) are calculated for situations where the elevator is stopping before the level of the floor. A speed reference limit vn _ιm for situations where the decel- eration reference would let the elevator advance past the floor level is calculated in block 126. If the speed refer¬ ence is below the limit thus calculated, the jerk will be as¬ signed a maximum value J4=J4 _max (blocks 128 and 130) and the procedure will continue with a renewed speed reference calcu- lation (Fig. 8) . The maximum value of the jerk, as well as its minimum value mentioned below, have been defined as pa¬ rameters for the elevator drive. If the speed reference is below the shortrun limit and the distance is above the shor- trun limit (block 132) , this means that it is no longer pos- sible to reach the floor level. In this case, the jerk value is calculated from the speed reference (block 134) and checked to ensure that it is not below the allowed minimum value J4 _mιrι or above the allowed maximum value J4 _max , and the jerk is assigned the value thus calculated, i.e. J4=j=a _de ² / (2*v _ref ) (blocks 136, 138 and 140) . If the calculated jerk is below the minimum value, the jerk will be assigned the minimum value J4=J4 _mn (block 142), or if the calculated jerk is above the maximum value, the jerk will be assigned the maximum value J4=J4 _max (block 150) .

When the elevator is stopping with normal deceleration, i.e. the limits in blocks 128 and 132 are not exceeded, the veloc¬ ity v (block 144) and distance d _a (block 146) are calculated using the speed reference and deceleration values. Next, a check is performed to see if the speed reference is below the velocity v and to ensure that the distance d to the floor

level corresponds to the calculated distance d _a closely enough (Δd = ±0.003 m) and that the flag has been reached. If the conditions are true, a value for the jerk will be calcu¬ lated from the deceleration reference and speed reference (block 152) . After this, a check is made to determine whether the calculated jerk is larger than the pre-selected value J _end , and if it is, then the calculated jerk will be accepted (blocks 154 and 156) . Otherwise the jerk will be assigned a zero value, in other words, the elevator will continue moving with constant deceleration (block 158). The procedure contin¬ ues again with the calculation of the next speed reference according to Fig. 8.

There are two limit conditions for distances too long or too short, and in addition there are conditions for normal situa¬ tions for the calculation of a final jerk. Before the limit is checked, the position checkpoint must have been reached. This ensures that the position data is accurate (corrected at the edge of the flag) .

In situations where the position data has not been updated, no flag has been detected, although according to the calcu¬ lated position data it should have been, the position error estimate produces a change in the deceleration a _di in ad- vance, which has an effect in the same direction as would re¬ sult when reaching the flag edge. But as the position error is taken into account in advance, the change is not as large as it would be without estimation.

It is obvious to a person skilled in the art that the embodi¬ ments of the invention are not limited to the embodiments de ^¬ scribed above, but that they can be varied within the scope of the following claims.